KEEL

Good day. Welcome to the Keel Infrastructure 1st quarter 2026 earnings conference call. At this time, all participants are in listen-only mode. After the speaker's presentation, there'll be a question and answer session. To ask a question, you will need to press star 11 on your touchtone telephone. Please note this call is being recorded. I would like to turn the call over to Jennifer Drew-Bear from Keel Investor Relations. Please go ahead.

Thank you, and welcome to Keel Infrastructure's first quarter 2026 conference call. With me on the call today are Director and Chief Executive Officer, Ben Gagnon, and Chief Financial Officer, Jonathan Muir. Before we begin, please note this call is being webcast with an accompanying slide deck. Today's press release and our presentation can be accessed on our website under the Investors section. Turning to slide 2. I'd like to remind everyone that certain forward-looking statements will be made during this call and that future results could differ from those implied in this statement. The forward-looking information is based on certain assumptions and is subject to risk and uncertainty. I invite you to consult Keel's 10-Q for a complete list, which will be available on our website and the SEC website.

Please note that references will be made to certain non-GAAP financial measures, and therefore may not be comparable to similar measures presented by other companies. We invite listeners to refer to today's press release and our file 10-Q for definitions on the aforementioned non-GAAP measures and their reconciliations to GAAP measures. Please note that all financial references are denominated in US dollars unless otherwise noted. Now, turning to slide 3. It is my pleasure to turn the call over to Ben Gagnon, member of the Keel Board of Directors and our Chief Executive Officer. Ben, please go ahead.

Good morning, everyone, and welcome to our first quarter 2026 earnings call. Today is a meaningful day for us. This is our first earnings call presenting as Keel Infrastructure. For those tracking this very closely, I want to take a moment to acknowledge what that represents. 2 years ago, we outlined a deliberate multi-year plan to transform this company. Wind down Bitcoin, build out our team, and reposition every megawatt we control towards the most significant infrastructure opportunity of our generation. That plan is now fully in motion. Since our last call just over a month ago, we have also completed our re-domiciliation to the U.S., officially rebranded as Keel Infrastructure, and closed the sale of our Paso Pe site. For those of you joining us for the first time, let me give you a clear picture of who Keel Infrastructure is and what we are building.

Keel Infrastructure is a North American digital infrastructure company. We own large-scale powered land sites across Pennsylvania, Quebec, and Washington that we are actively developing into over 2 gigawatts of high-performance computing campuses for lease to investment-grade hyperscalers, neo clouds, enterprise, and government clients. The Keel name captures who and what we are. The keel is the structural backbone of a ship, unseen but essential, converting energy into forward motion. That is exactly what we do for our tenants. We enable and accelerate the data center growth that makes tomorrow's economy possible. Turning to slide 4. Let me take a step back now and talk about why we are attracting so much attention from potential tenants and why we're set up to create tremendous value for customers. The conversation in HPC and AI infrastructure has shifted fundamentally over the past 12 months.

Customers are not asking, "Can you build data centers?" They are asking, "When can you deliver power in the right location on a timeline that actually matters to my deployment schedule?" "How are you ensuring you can deliver?" The answer to those questions is what separates sites that get leased from sites that sit empty. Our strategy is customer-centric and is structured around solving their highest value constraints. 1, short timelines to power. Our sites have secured power available starting in 2027, enabling customers to accelerate deployment relative to building out interconnections organically. In PJM, Quebec, and Washington, a new large load interconnection can take between 4-10 years. We have already done that work. That timeline advantage is not incremental. It is transformational for customers trying to deploy compute at scale. 2, prime locations.

Panther Creek, our flagship campus, is a great example of the value our locations bring. The site sits 2 hours away from Philadelphia and New York in the PJM energy market, surrounded by established hyperscaler and neo cloud data center infrastructure. Our other campuses follow the same principle, proximity to metro areas and surrounded by our customers' established infrastructure. These are not secondary energy markets. These are primary markets where our customers are actively trying to expand and finding that supply at this time does not exist. 3, a proven permitting strategy built on transparent stakeholder relations. While strong community engagement and support has always been a pillar of our culture at Keel, recent headlines are reinforcing just how critical this is. Our permitting team has decades of regional experience, and we proactively build genuine relationships with the communities around our sites. That approach produces results.

Zoning is now complete at all three near-term sites. Land development and environmental permits are on track, including our preliminary land development approval at Sharon. Customers who have watched other developers miss permit milestones appreciate what this means for Keel's execution certainty. 4, proven delivery partners with hyperscaler grade track records. With power, land, and community support, we have the foundation in place for success. Customer confidence ultimately comes from execution, which is why we've built a partner ecosystem designed to deliver that certainty. Working with Turner Construction, Corgan, Vertiv, and T5, our customers do not need to take development execution risk on an untested team. Potential customers are looking at our construction and engineering partner roster and seeing our collaboration with best-in-class infrastructure and construction partners that have demonstrated experience delivering for hyperscalers. 5, future-proof designs.

We are advancing architecture and engineering in parallel with customer conversations, which means that when a customer is ready to commit, we will be ready to easily adapt to their final specifications. We are also thinking ahead. With rapidly evolving technology, it has never been more critical to future-proof our data center development. We are thinking about our customer needs in 2027 and beyond, not just what they need now. Customers value that. Turning to slide 5. Our portfolio is focused on high barrier to entry markets in Pennsylvania, Washington, and Quebec. In these markets, our ability to accelerate timelines and enable regional growth creates real value for customers. Our 2026 priority is clear. Sign 3 leases by year-end, 1 at Panther Creek, 1 at Sharon, and 1 at Moses Lake. We have the right power in the right places with the right timelines.

As Jonathan will walk through, we are better capitalized than at any point in this company's history, with more than enough liquidity to advance all 3 sites through permitting and lease execution. Across all 3 of our near-term development sites, we are running 3 work streams simultaneously. Finalizing permits, advancing A&E aligned with customer specifications, and actively commercializing to secure highly financiable leases with investment-grade tenants. That parallel execution model is intentional. In this market, customers are making site decisions now. They are looking for partners who can show them a clear, credible path to power, we create that visibility by working with great partners and advancing all 3 work streams together. When a customer is ready to commit, we are ready to build. Let me take you through each of our 3 near-term sites. Turning to slide 6.

Starting with Panther Creek, our flagship campus in Eastern Pennsylvania and the centerpiece of our near-term development plan. We have 350 MW of secured gross capacity with PPL under an ESA. Development is structured in phases, with an expected ready for service date in 2027 and additional expansion capacity beyond that. Permitting is a subject I know investors track closely. Let me walk through our approach with precision. Permits fall into 3 broad categories: zoning, development, and environmental. Full permitting requires completion across all 3. Our execution strategy is built around local expertise and proactive engagement, planning, and transparency.

We have assembled a team with deep regional knowledge anchored by a head of permitting with decades of Pennsylvania experience, and that local presence allows us to move efficiently through jurisdictional requirements, and just as importantly, to engage productively with the communities around these sites, who are always key partners in Keel developments. On the permitting progress. Zoning approvals were completed in February, including the data center ordinance approval by the Nesquehoning Borough, a meaningful community milestone. Land development and environmental permits remain in process and are on track. With zoning secured and a clear line of sight on development timelines, we are active in commercialization. To be clear, we do not need to wait, nor are we waiting for every permit to negotiate leases. We give customers the visibility they need to make decisions and the certainty that they need to commit.

In terms of the customer profile for the site, the scale and location of Panther Creek positions it squarely for hyperscalers and the largest neo cloud operators. 2 hours from New York City with 8 fiber metro networks within 10 miles and direct proximity to established data center clusters, this is the kind of site that gets on a short list quickly. We are in active conversations with multiple potential customers, and the engagement quality has been strong. Beyond the 350 MW of secure power at this campus, we are currently evaluating the conversion of our existing 60 MW ISA to firm service, which could bring total gross capacity upwards of 400 MW or 430 MW. A new load study conducted in 2025 supports potential expansion beyond 500 MW for the overall campus over the longer term.

We will provide updates as that conversion evaluation progresses. The point is, Panther Creek is a unique asset. It has the proximity and scale to service East Coast inference and training markets for years to come. Turning to slide 7. Moving to Western P.A. We have 110 MW secured via an ESA with FirstEnergy. A 30 MW substation is operational today with an additional 80 MW substation under development. Sharon received full zoning permits last month. That is a significant milestone, and it gives customers increasing confidence in our delivery timeline. Land development has been preliminarily approved, and environmental permits are in progress and on track.

This site is actively being commercialized with an expected ready for service date as early as 2027. Sharon sits within the PJM market with strong fiber infrastructure across 9 metro networks within 10 miles in proximity to Pittsburgh and Cleveland, two markets that are underserved relative to the East Coast. In terms of customer profile, the capacity and location make Sharon a strong fit for a hyperscaler, neo cloud operator, or large enterprise customers looking to establish a position in Western PJM. We are in active conversations with multiple potential customers, and the response to our programming progress has been positive. Turning to slide 8. Finally, Moses Lake, our 18 MW site in Washington State.

Small but mighty, Moses Lake is located adjacent to one of the most proven data center markets in the U.S., the Quincy, Washington corridor, which has been home to hyperscaler infrastructure for nearly two decades. Power availability in this region has become one of the most constrained in the country. The combination of existing cluster density and tightening power supply means that operators who need megawatts here have very limited options to grow organically. We are one of those options to establish a footprint or expand an already established operation. Moses is the only site where we made a deliberate capital decision ahead of commercialization. We purchased critical modular data center equipment in advance. That decision enables us to offer customers an accelerated deployment timeline that is not available through a traditional stick build approach. Speed matters to our customers, and we engineered our deployment model to deliver it.

Zoning in Moses is complete. Land development and environmental permits are in progress and on track, and the Bitcoin mining operations are actively being decommissioned. Like our Pennsylvania sites, Moses Lake is actively being commercialized with strong inbound interest and ongoing engagement with multiple counterparties. In terms of customer profile, the scale of the site positions it as an ideal fit for emerging neo clouds, enterprise, and government customers who need fast, reliable access to the Pacific Northwest market and do not require a campus scale commitment to do so. Faster timeline, smaller megawatt commitment, right market. That is a compelling combination. Across all 3 sites, we have clear line of sight to full permitting, active commercialization, and tangible momentum towards signed leases in 2026. We look forward to keeping everyone updated on our progress. Turning to slide 9.

From a value creation standpoint, a signed lease is the single most important inflection point for our business. A signed lease does 3 things. It converts our development assets into long-term contracted cash flows. It unlocks access to low cost, non-dilutive project financing. It significantly reduces execution risk for every stakeholder in our capital structure. There is a reason we are intensely focused on getting 3 leases signed this year, where we expect each lease to be an event that reshapes how this company is valued. We are executing against all 3 simultaneously right now. The second value driver we are executing this year is to increase our secured capacity from both expansion capacity and new organic growth opportunities. The 3rd value driver will be delivering on megawatts in 2027.

We believe that these three inflection points are key drivers of value creation for our shareholders in the near term and long term. With that, I'll turn it over to Jonathan to walk through our financial position and strategy.

Thanks, Ben. Turning to slide 10, I want to open with a simple message. We are better capitalized today than at any point in this company's history, and our liquidity position gives us something invaluable in this market, the ability to both advance and de-risk our sites at the pace our customers require, and to make commercial decisions from a position of strength, not necessity. As discussed during our last call, our financial strategy rests on 3 principles: capital allocation, capital formation, and capital structure. Each directly supports our ability to execute our goal of signing 3 leases this year. Before I walk you through our strategy in more detail, I'll briefly go over our results for the quarter. Turning to slide 11. As a reminder, as of Q3 2025, the Paso Pe facility in Paraguay has been classified as held for sale.

As a result, all revenues, operating costs, and asset balances associated with Paso Pe are treated as discontinued operations in our Q1 2026 financials. When I refer to continuing operations, I am speaking exclusively about our North American platform, which is the foundation of all our transition into HPC and AI infrastructure. With that, revenue for first quarter 2026 was $37 million, down 23% year-over-year. Operating loss for the quarter was $98 million, including non-cash depreciation of $28 million compared to an operating loss of $35 million in Q1 2025, which included $18 million of non-cash depreciation. The year-over-year change primarily reflects a $41 million loss related to change in fair value of digital assets in Q1 2026, compared to a loss of $23 million in Q1 2025.

Loss from continuing operations was $128 million or a $0.21 loss per basic and diluted share, compared to a loss of $38 million or a $0.08 loss per basic and diluted share in Q1 2025. The changes reflect the increase in operating loss and a $22 million loss from the extinguishment of the Macquarie credit facility in Q1 2026. For the first quarter 2026, our adjusted EBITDA was negative $17 million, down from $7 million in 2025. The difference was largely due to an increase in energy and infrastructure expenses of $15 million and an unfavorable change of $7 million in the gain or loss from the sale of digital assets. Turning to slide 12. Let me turn to our capital position.

Since our last call, we have taken two actions that further strengthen our balance sheet. First, we closed the sale of our Paso Pe site, which brought forward roughly 2 to 3 years of estimated cash flow under current market conditions in cash and upfront. Second, we have continued to actively manage our Bitcoin holdings, selling into strength and methodically converting a volatile asset into the stable capital our development business requires. During the period beginning January 1, 2026, and ending May 8, 2026, we sold 269 Bitcoin for $20 million in proceeds as part of our previously communicated plans to sell our Bitcoin holdings in 2026. Current liquidity as of May 8, 2026, stood at approximately $533 million in cash and Bitcoin. Let me put that number into context.

This fully funds the capital required to advance Panther Creek, Sharon, and Moses Lake through lease execution, as well as the start of construction at Moses Lake and covers our G&A through 2028. We believe this liquidity is a strategic advantage. We can continue developing at the speed our customers require while maintaining discipline and deploying capital where the returns are most compelling. Let me now walk through the three principles that guide our financial strategy. First, capital allocation. Every dollar we are deploying today is advancing our three priority sites toward lease execution. We believe it is the highest return use of capital available to us at this stage of the company's development. Second, capital formation. As I noted, we have the liquidity to reach lease execution across all three sites without the need to tap into debt or equity capital markets.

That said, we will remain opportunistic if attractive opportunities arise. Once we execute leases, we would expect to transition to project-level financing models supported by long-term contracted cash flows, enabling us to fund construction with a high proportion of non-recourse capital while preserving flexibility at the corporate level. The institutional financing market for HPC AI infrastructure continues to strengthen, we believe we're well-positioned to access it on favorable terms at the appropriate time. Third, capital structure. We operate with a disciplined liquidity strategy so that we can remain flexible when making commercial decisions. As I mentioned a few moments ago, we have more than adequate liquidity today to execute against our strategy without the need to tap into capital markets.

That said, we'll always take the necessary steps to ensure a strong balance sheet, and we would envision having a credit line and/or an ATM in place at some point this year, as we believe these are prudent tools for any public company to have available. Again, liquidity and capital strength are directly supportive of our commercial strategy.

Thanks, Jonathan. Before we open for questions, I want to drive home a few things. This company has done what it said it would do. We said we would build a North American infrastructure platform. We built it. We said we would exit Latin American megawatts. Done. We said we would re-domicile to the United States and rebrand. Complete. We said we would position our megawatts in the most capacity-constrained, high-demand markets in North America, and this is exactly where 100% of our portfolio sits today. The case for Keel Infrastructure is direct. Power availability is the single biggest bottleneck constraining the growth of the AI economy. We control scarce, deliverable power in 3 of the most supply-constrained markets in North America, allowing us to work alongside our customers to solve that challenge together.

We have the sites, the team, the permits in progress, the partners, and the balance sheet to execute. We are executing now. 3 leases signed by year-end, revenue commencing in 2027. That is the plan. That is what we are focused on delivering. I want to close by acknowledging our fantastic team. The pace and the precision with which we have executed this transformation, the transactions, the hires, the permitting progress, the commercialization, is not the result of any 1 decision. It is the result of hundreds of well-made decisions by a team that is fully committed to this mission. I've never had more confidence in our team and our ability to deliver. I look forward to continuing to update you on our progress. With that, I would like to open the call to Q&A. Operator, please go ahead.

Thank you. As a reminder, to ask a question, please press star 11. If your question has been answered and you'd like to remove yourself from the queue, please press star 11 again. Our first question comes from Mike Grondahl with Northland. Your line is open.

Hey, thanks, guys. Ben, maybe specifically on Sherbrooke, you'd kind of talked about hyperscaler customers, neo clouds, and large enterprises. Can you talk a little bit about the pros and cons or the terms from each category and what metrics you're going to use to decide on a lease?

Thanks, Mike, it's a great question. When you're looking at all the different available potential tenants for these sites, there's obviously going to be pros and cons across the various categories. I think broadly speaking, you know, what you see from a hyperscaler client is probably, you know, a little bit tighter on the economics. That's largely offset by the quality of the credit and the confidence in the long-term contract there. NeoClouds are generally paying a bit of a higher price. They also come with a higher cost to capital. There's a balancing act.

For us, really, it's about finding the right balancing act between the counterparty, the economics of the contract, and the cost to capital, but not specifically trying to get a hyperscaler over a NeoCloud, but really trying to optimize across those three variables.

Any sense where you're leaning today?

I don't wanna get into, you know, exactly where we're gonna go, but on the slides, what we did indicate for each site was the potential, kind of tenant profiles. That should give you an indication of, kind of where we're leaning for each site because most of the site's scale is determining the kind of customer demand that we're receiving.

Got it. Then just lastly, how has demand changed over the last 90 days?

I don't think it has changed, Mike. It's still present. It's still incredibly strong. There is some emerging, you know, questions around kind of global investments in HPC and AI versus the U.S., given what's happened in the Middle East and given the geopolitical uncertainty of investing everywhere else. I don't think we've seen a real change in demand. It's more or less a reinforcement of what was already there before the conflict, a preference to invest in the United States. Now we're seeing, you know, just a much stronger reinforcement of that. I think demand is as strong as it was, you know, 90 days ago or 120 days ago.

Fair. Okay. Thank you.

Thanks, Mike.

Thank you. Our next question comes from Brett Knoblauch with Cantor Fitzgerald. Your line is open.

Hi, guys. Thank you for taking my question. On Panther Creek, which seems to kind of be, like, the largest initial site for you guys or the flagship site. I didn't know the slide that says we're kind of waiting on, you know, environmental and land. Could you maybe just help with a timeline on that? You know, is that still a 3 Q, you know, event? Could it happen sooner? Is that absolutely necessary, call it, to happen pre-lease execution?

It's great questions, Brett. You know, we're still tracking on the exact same timeline that we indicated on the last Q4 call a couple of weeks ago, which is kind of a mid-late summer timeframe. This is what we're lining up for right now. You know, what we, you know, wanna make clear in terms of the process is lease negotiations and permitting are as a parallel process. It's not as if you need those in hand to begin a successful lease negotiation, but you have to be able to show a very confident and credible pathway, with a high confidence that you'll achieve it on the timelines you're gonna achieve it to be successful in those lease negotiations.

We achieved that earlier this year, which is why we've been active in the commercialization strategy across all 3 of those different sites. We shouldn't expect that, you know, the timing of the permits is going to have, you know, a slowdown in terms of the lease execution. Those are simultaneous, and we would be looking to complete the permits before executing the final lease. The negotiation and the permit applications continue in parallel.

Awesome. Maybe just as a follow-up, you know, I think what we're hearing across most of the space is that kind of capacity for 2026 is sold out, anything with an ARPA state in 2027 should be relatively attractive. Then you guys are also designing, at least sharing for Vera Rubin. Are you seeing any, you know, change in conversation given it's a Vera Rubin kind of design, relative to maybe other sites that might be maybe Blackwell? I'm just curious if you're seeing, like, an uptick in demand for what would be a Vera Rubin site.

The Vera Rubin technology is very different than Blackwell. The engineering requirements are a magnitude of order more complex and sophisticated than Blackwell. The conversations are relatively different. I think in terms of Blackwell, nobody's actually received their first allotment. Or sorry, in terms of Vera Rubin, nobody's actually received their first deliveries of Vera Rubin. You know, the conversation with Vera Rubin is much more about planning for the future and are trying to accommodate for the equipment that is really just kind of coming off the first lines of the production run right now. Whereas Blackwell is more of a known technology and a known engineering standpoint. I would say from a demand perspective, we see more demand for Vera Rubin with our timelines of 27.

The biggest difference in the conversation is really just the changing in real-time engineering requirements from Nvidia for the Vera Rubin technology stack because this is just starting to emerge in the market now.

Awesome. Thank you, Ben.

Thanks, Brett.

Thank you. Our next question comes from Bill Papanastasiou with Chardan Capital Markets. Your line is open.

Yeah, good morning. Thanks for taking my questions. Previously, I believe management mentioned the timelines for clearing permitting would be mid to late summer. Not sure if this was mentioned on the call, how is that trending? Has that timeline shifted at all now that you have zoning at all three sites?

Hey, Bill. Thanks for the question. Yeah, we mentioned that on the Q4 call. Since we've had the Q4 call, we've cleared out on a few more permits, including zoning and preliminary land development at Sharon. Everything is tracking according to our plan. We still have high confidence on a mid to late summer timeframe across those 3 sites. You know, to permitting, obviously things can go a little bit faster or a little bit slower, we've got high confidence on those timelines.

Thanks. Can you just speak to your Bitcoin mining operations, where steady state today, I believe in Q4 was around 14 exahash? How should we think about that throughout the remainder of the year? Thanks.

Yeah, it's still around 14 exahash. It should continue to trickle downward over time. Right now, the Washington site is being decommissioned. That's our first U.S. site where we've actively decommissioned Bitcoin mining. Before, it was all coming out of Latin America. As we break ground and work on development across Panther Creek and Sharon, we will also be decommissioning Bitcoin mining at those sites. We're gonna try and line up the Bitcoin mining decommissioning as best as possible with the construction schedule and mining economics so that we can try and optimize and maximize the capture of the value and the cash flows there. We'll continue to provide an update to the market as we move forward throughout the year, Bill.

You should expect it to trickle down from 14 to probably somewhere around, I would think, 5 exahash around the end of the year.

Perfect. Thank you.

Thank you. Our next question comes from Michael Donovan with Compass Point. Your line is open.

Hi, guys. Thanks for taking my questions. On Moses Lake, the slide deck states there is a secured option to acquire neighboring property with additional capacity. Can you size the potential expansion opportunity beyond the current 18 MW? What needs to happen for that option to move forward?

We have a secured option for an additional 10 MW in the area. Nothing really needs to happen other than our desire to exercise the option. The power is there, it's secure, the land is there, the due diligence is done. Really, it's just about us wanting to exercise the option. When you go out and you do market for these sites, one of the strategic features to have in these conversations is not only to have secured power today, but to have the ability to expand that infrastructure and expand that capacity over time. Securing the option, you know, as of right now is this great marketing benefit for us when we're going through the commercialization strategy that gives us and the customers a potential to continue to scale up in that, in that region.

Thanks, Ben. Also on question, can you unpack the scope of the May third purchase commitment and clarify whether all major long lead equipment has been acquired?

We've secured basically everything that we need to do for the site with regards to the modular infrastructure from Vertiv, the transformers and the backup gens. Last thing that we really needed was the backup gens, which is the last thing that we had secured. Moses Lake is, it's got all of its equipment that it needs for its development. There's a few odds and ends, but all of the key critical pieces have been secured.

Appreciate that, and congrats on progress.

Thank you.

Thank you. Our next question comes from Martin Toner with ATB Cormark. Your line is open.

Great question. Congrats on all your progress. SG&A ticked up this quarter. Can you maybe talk to what we can expect for the rest of the year and just in general, maybe, you know?

Good morning, Martin. It's Jonathan. How are you? Could you repeat the back half of your question? I did hear you ask about expectations for SG&A for the remainder of the year and missed a bit at the end.

Yeah, just talk a little bit about what investment that increase in SG&A represents.

Thank you. That's very clear. We'd expect our run rate cash SG&A to run about $25 million a quarter for $100 million a year, ±. At the SG&A level, we've got a number of offsetting factors related on the one hand to the wind down of elements of the Bitcoin business, and then on the other hand, adding specialized expertise in respect of the HPC AI data center build-out.

Perfect. Can you talk a little bit about capacity at the site?

It was a little hard to hear that, Martin, but I believe the question was just an update on Quebec site and Sherbrooke. Is that correct?

Yes, please.

We continue to make good progress with our 96 megawatt campus in Sherbrooke. We're hoping to have an update on today's call, but we should have an update on the Q2 call, which would include, you know, our plans for consolidating our 3 Bitcoin mining sites in Sherbrooke, our 48 megawatt bunker site, as well as our 30 and our 18 megawatt sites, Leger and Garlock. To a single 96 megawatt site in the same town. We're continuing to progress those conversations with the City of Sherbrooke and Hydro-Sherbrooke. We have high confidence that we're going to be able to get all of those, you know, I's dotted and T's crossed to wrap this up and to be able to provide our plans to the public.

We're getting quite excited about our plans in Sherbrooke. We think that it represents one of the few permitted HPC AI campuses in Quebec that will be under construction in the near term.

Great. Thank you very much. That's all for me.

Thank you. As a reminder, to ask a question, please press star 11.

Thanks very much. Good morning. Thanks for the positive commentary on the demand environment. Do you think you could maybe give us a little bit of color on what you see as the biggest gating factors for your growth over the next few years, and if there are any, you know, long lead time obstacles that you're trying to overcome?

I think the biggest gating factor, Brian, is just bandwidth, to be honest with you. We've built a great team. We're continuing to build a great team, but we have 2 GW worth of development pipeline to execute against. There's a tremendous amount of technical details and complexity associated with these projects. We've done a great job in terms of increasing our bandwidth with adding more people, selling off non-core assets, you know, completing these structural things, which really help to simplify the business and the administration of the business, like re-doming off to the U.S. and completing our pivot out of Canada and LATAM. All of that stuff is adding into that.

We've also had a lot of success with, you know, early integrations of AI into people's workflows and to people's work streams, which is helping productivity as well. I think that's probably just the biggest constraint is bandwidth. You know, that's something that we're continuing to improve upon as we continue to add people to the team, continue to add great partners like Turner Construction and Corgan on A&E and all these other different areas. I think we've got a very good pathway to address those and to execute across all of our different campuses.

Great. Thanks for the color.

Thanks, Brian.

Thank you. Our next question comes from Mike Colonnese with H.C. Wainwright & Co. Your line is open.

Hi. Good morning, Ben and team. Thank you for taking my question. Just one for me today. If you could just talk about the pricing dynamics that you're seeing from negotiations with prospective tenants here. Is it fair to assume that Keel could secure better economics on a lease than what we've seen in the marketplace recently, specifically given the location of your sites in PJM and Washington, then paired with your data center design, which sounds like it's aiming to support the Vera Rubin deployments?

Thanks, Mike. It's one of the questions that we're paying very, very close attention to. It's one of the things that we've been talking about for some time now, that we believe that the economics are continuing to improve as the scarcity continues to get worse and demand continues to accelerate. I don't wanna get locked in on any sort of fixed numbers with lease economics. I think the broad trend is quite clear. I don't think it's changed or slowed down at all. You know, the market demand for this growth is very, very high. We're seeing hyperscalers reconfirm their commitments, in some case increase their commitments, in some cases, you know, making pretty loud statements on quarterly calls around the opportunity cost of the missed revenue for not having that compute in place.

You know, we do think that this is probably going to be a trend that continues to play out for years to come. We look forward to, you know, taking advantage of our energy position in an increasingly energy-constrained market.

Very helpful, Ben. If I could just squeeze one more in, actually. On the CapEx side, as you've gotten further along in your basis of design, with your various campuses, has your capital requirements or CapEx deployment needs changed at all since your initial framework, when it comes to deploying these data centers?

Hi, it's Jonathan. Generally speaking, no. Our views on CapEx deployment have not changed since our initial framework. We're comfortable with our current plans, and people always ask about guidance on this topic, and we'd say the figures generally used as rule of thumb throughout the industry should be fine as a practical matter.

Great. Thanks, guys.

Thanks, Mike.

Thank you. Our next question comes from Nick Giles with B. Riley Securities. Your line is open.

Thanks, operator. Good morning, guys.

Good morning.

Much of today's discussion has centered on your first 3 sites. I wanted to ask about Scrub Grass. Can you just give us a sense for progress there specifically? What do you see as the key milestones for that site over the, you know, next 6 to 12 months?

Thanks, Nick, and I appreciate your enthusiasm for Scrub Grass, which is an enthusiasm that I share. You know, I find Scrub Grass to be a really exciting project for us. It's likely going to be the crown jewel of the company in the coming years, but there's still a lot of work for us to execute against before it can achieve that kind of status. The reality is that this is gonna be one of the largest data center campuses in Pennsylvania. But we've got to get power secured from a couple of different angles. And it's just gonna take some more time to do that. You know, on the grid connection side, the detailed load study is continuing forward.

We should expect to have, you know, a indication as to what the results of that are sometime around, you know, the very end of the year in Q4. We're working on securing the energy pipeline lateral construction and the energy contracts as well as the agreements with either an IPP or, you know, a similar firm to come out and deploy nat gas turbines on site, even evaluating options for us to do it ourselves. It's a little too early to really say exactly what's gonna happen or when it's gonna happen, but we do share your enthusiasm for that site and its potential. We do think it's gonna be one of the more transformative value creation opportunities for the business and for shareholders.

It is one of our big focuses for the company and for management this year, is to secure the megawatts at Scrub Grass and pull them out of that expansion bucket into the secure, secured bucket. That would more than double our secured capacity by doing so, and would give us a real powerful giga-campus in Pennsylvania. If I could just build on that for one, you know, brief moment. What we've seen in the market is that the giga-campuses are fiercely contested. Especially if you have a giga-campus outside of Texas, which are increasingly rare. Those sites have a more competitive, tension-filled process when they're going through the commercialization stage. We would look forward to taking full advantage of that in a capacity-constrained market.

Thanks for that detail, Ben. That's super helpful. Just to clarify, how much power does the detailed load study cover?

The detailed load study is for 750 MW.

Got it. Great. Thanks for all the color today. Continue. Best of luck.

Thank you, Nick.

Thank you. I'm showing no further questions at this time. I'd like to turn the call over to Ben Gagnon, CEO, for closing remarks.

Thank you everyone for attending our Q1 call. At this time, we'll go ahead and end the call, but we'll continue to provide updates for you on our website and through the normal investor channels. Thank you.

Thank you for your participation. You may now disconnect. Everyone, have a great day.

Good day, and welcome to the Bitfarms' Fiscal 2025 conference call. At this time, all participants are in listen-only mode. After the speaker's presentation, there'll be a question and answer session. Instructions will be given at that time. Please note this call is being recorded. I would like to turn the call over to Jennifer Drew-Bear from Bitfarms Investor Relations. Please go ahead.

Thank you, and welcome to Bitfarms' Fiscal Year 2025 conference call. With me on the call today are Ben Gagnon, Chief Executive Officer and Director, and Jonathan Mir, Chief Financial Officer. Before we begin, please note this call is being webcast with an accompanying slide presentation. Today's press release and our presentation can be accessed on our website under the investor section. Turning to slide two. I'd like to remind everyone that certain forward-looking statements will be made during the call, and that future results could differ from those implied in this statement. The forward-looking information is based on certain assumptions and is subject to risks and uncertainties. I invite you to consult Bitfarms' 10-K for a complete list. Also, please note that references will be made to certain non-GAAP financial measures and therefore may not be comparable to similar measures presented by other companies.

We invite listeners to refer to today's press release and our 10-K for definitions of the aforementioned non-GAAP measures and their reconciliations to GAAP measures. Please note that all financial references are denominated in U.S. dollars unless otherwise noted. Now turning to slide three, it is my pleasure to turn over the call to Ben Gagnon, Director and Chief Executive Officer. Ben, the floor is yours.

Good morning, everyone, and welcome to our Fiscal Year 2025 earnings call. In 2025, we made a bold decision to walk away from our legacy business, Bitcoin, and build the infrastructure in North America for what comes next, HPC and AI. It was a year of deliberate and consequential transformation with a clear mandate, secure North American pipeline, strengthen our balance sheet, accelerate site development, and position ourselves to engage customers from a place of operational momentum at the peak of the energy bottleneck constraining the growth of AI. I can say with confidence and pride that we accomplished exactly what we set out to do. The foundation you see today, the capital structure, the sites, the team, the strategy, was engineered through deliberate choices, developed with discipline, and built to propel us forward.

We made foundational changes to reposition the business and made 100% of our focus on North American HPC infrastructure development. No half-measures, no compromises, and in time, no Bitcoin. We built a new company, and while we are presenting as Bitfarms today, tomorrow marks our beginning as Keel Infrastructure. The name says it all. A keel is the bottommost structural component of a vessel. It's what keeps it stable and moving forward in the right direction, regardless of the condition above the waterline. It is structural, it is essential, and it is exactly how we see our role in the HPC and infrastructure landscape. We are not here to compete with hyperscalers or Neoclouds. We are here to enable them. Our focus is providing the critical and largely invisible foundation that will allow the world's most advanced AI platforms to deploy on time and scale without interruption.

We expect to close the re-domiciliation and finalize our rebranding efforts tomorrow, April 1st, and will begin trading under the ticker Keel two business days after completion of the transaction on the Nasdaq and the TSX. We are entering this new phase from a position of strength. With over 2 GW in our pipeline, Keel is a regional leader with some of the largest power land portfolios in some of the highest demand markets in North America, and with robust financial strength to execute against our plan. Our current liquidity is far in excess of the CapEx budgeted to get us through permitting and ultimately to start signing leases, giving the company significant financial flexibility to execute on our strategy. Our strategy is equally as clear. We are designing all of our site and campus developments as either powered shell or colocation facilities.

We believe this is where we can deliver the most value to shareholders and serve our potential customers at the speed and to the specifications they need. We were originally exploring, in parallel to colocation, the potential benefits of pursuing a small amount of GPU-as-a-service at our Washington site, Moses Lake, where due to the lowest cost power for data centers in the country and a relatively smaller footprint, we believed it could be an avenue to drive additional shareholder value. Since our last quarterly call, we have spoken with an increased volume of potential customers. It's clear from those conversations, the most accretive business model for the site is one of colocation. This is not specific to Moses Lake and applies to all of our other sites as well, where demand is even higher.

We will focus on what we do best, being an infrastructure developer and owner. This plays directly to our core competencies. We are a team of developers united by disciplined action, building cost-effective, institutional-grade infrastructure at the pace our customers require. The same capabilities that built our energy platform, speed to market, capital discipline, operational rigor, are precisely what HPC and AI deployments demand today. This is just the natural extension of what we do best. With all the pieces in place and with the overwhelming support of our shareholders who voted over 99% in favor of the HPC and AI pivot, the U.S. re-domiciliation, and the rebrand, starting tomorrow, we are Keel Infrastructure. Turning to slide four.

When we set out on our pivot, we developed a three-year transformation plan, one that as of today, we are nearly halfway through completing. In 2025, we did the intensive foundational work for our transformation, including the Stronghold acquisition, securing more power in Pennsylvania, rebalancing the portfolio to North America, $588 million raise, fully institutional and oversubscribed, our U.S. GAAP transition, New York headquarters, and establishing a new executive team. This work is done. With power and land secured in some of the power markets that matter most, a team of internal experts and strategic partners that have built data centers for the largest companies in the world, and a balance sheet engineered to see us through 2026. We are well-positioned to continue our site development and deliver against the timelines our prospective hyperscalers and Neocloud customers need. 2026 is all about execution.

Effective tomorrow, we will have completed our re-domiciliation to the United States and officially rebranded as Keel Infrastructure. Two major milestones that position the company for the next phase of growth. With that complete, we expect the next significant milestones to come from executing against our development at Panther Creek, Sharon, and Moses Lake, where we are moving full steam ahead and working diligently across three simultaneous and active work streams. One, finalizing permits, which we expect to be done in the coming months. Two, continued work on architecture and engineering in line with ongoing customer conversations and requirements. Of course, three, our go-to-market to secure highly financiable leases with investment-grade tenants. Commercialization is well underway.

The upcoming milestones investors can expect are completion of pre-construction activities like permitting, progress in customer engagement, and ultimately lease execution, which we are confident we can achieve this year and will be major catalysts. 2026 is also the year where we expect to leave Bitcoin and Bitcoin mining behind. While we are probably one of the first miners to commence wind down of our Bitcoin mining exposure to reinvest that capital into infrastructure for HPC and AI, we will be accelerating those efforts in 2026 as site developments progress. 2027 is all about delivery. This is the year when we anticipate that sites would come online, we begin delivering megawatts to customers, HPC and AI revenue really begins, and we complete our transition to a premier North American HPC and AI infrastructure company.

By the end of 2027, we expect Keel will be a proven infrastructure developer and a regional leader across Pennsylvania, Washington, and Quebec, and we will just continue to grow and scale from there in 2028 and beyond to over 2 GW as we execute against our expansion capacity. Turning to slide five. In HPC infrastructure, power, location, and timelines are everything. We hold something scarce and valuable. Secured power, land, and expansion capacity in Pennsylvania, Washington State, and Quebec. Some of the most in-demand markets with some of the biggest barriers to entry. We know it, and so do our potential tenants. Our campuses offer solutions to hyperscalers and Neoclouds' greatest scaling problems. Location, proximity and fiber connectivity to major metro areas and data center clusters, solving for latency issues and giving our tenants proximity to their own customers and other data centers.

Timelines, a robust secured power for 2026, 2027, and with expansion capacity in 2028 is highly coveted in an environment where energy capacity is hard to find and multi-year wait lists are the norms. We create value for tenants by enabling them to deploy years earlier by leasing from us rather than to invest in growing organically. An energy-efficient, cool climates. The lower the PUE, the more critical megawatts. Panther Creek is a great example of seeing the hyperscaler and Neocloud's appetite at play. While there was a lot of interest in the site last year, inbound customer activity surged after we secured zoning in February. This is not a coincidence. It is a proof point and one that we've been making for the last year, but may still be confusing to some investors.

We'd like to be clear that investment-grade tenants value de-risk sites where they can move from lease to revenue fast. The more we advance, the better our leverage. The better our leverage, the better the leases, and the more long-term value we create for shareholders. Turning to slide six. It is indisputable that power is the binding constraint for AI infrastructure deployment and will remain so for the coming years. Leading investment banks, Goldman Sachs, JPMorgan, Wells Fargo, Guggenheim, Moelis, they've all published extensively on this. The consensus is clear. New power generation cannot come online fast enough to meet AI demand today, tomorrow, or in the next five years. This bottleneck is structural, not cyclical. Hyperscalers and Neoclouds that used to plan on 12 month horizons are now locking in 24 to 36 month supply chain commitments.

Not tied to specific projects, but as platform-level agreements, and are now actively competing for the power and land to deploy it. While you are probably familiar with this information, here you can see a summary of the five development sites, the power we have secured, and in some cases, the incremental power opportunities that make up our 2.2 GW pipeline. Turning to slide seven. I want to take a moment to put our current valuation in context, because there is a meaningful disconnect between where we trade today and the value we are positioned to capture as a company. When we analyze our current valuation against our peers, the picture becomes clear.

At approximately $1.9 million per available megawatt of secure 2027 capacity, we're trading in the middle of a Bitcoin miner group, valued at roughly $1.7 million-$2.1 million per 2027 megawatt. Meaning we are being valued based on having power, but not what we are doing with it. For shareholders and bondholders, we see three distinct catalysts, each capable of driving meaningful re-ratings. The first is obviously lease execution. Across our sector, companies that have signed leases trade at $4 million-$6 million per 2027 megawatt. A two to three times premium to where we are today. This is the market's consistent signal, driven entirely by lease execution. Not facility delivery, not revenue generation, just signed leases. A signed lease secures revenue and financing, de-risking the developments. The market pays for that.

With nearly 500 MW actively being commercialized today and visibility on permitting across Panther Creek, Sharon, and Moses Lake, this catalyst is well within reach. The second catalyst, and arguably the most powerful for long-term holders, is securing our expansion capacity. Two-thirds of our 2.2 GW portfolio, or approximately 1.5 GW, is expansion capacity, which we believe the market is assigning little to no value. While securing these megawatts is a process that will take more time, we believe additional megawatts can be secured in the second half of 2026, requiring very little CapEx, but representing significant embedded value as powered land even before a lease is signed or there's a shovel in the ground. The third catalyst is delivering in 2027.

Once facilities are de-risked through commissioning and begin generating revenue under long-term contracts, the development risk should drop dramatically and the operator valuation numbers become transformational yet again. We are not taking a leap of faith on technology, our ability to steer power or market demand. The tech is here. The power is secured. The sites are advancing. The inbound demand is real. What the market has not yet priced in is the transformation that happens when a developer becomes a counterparty when we move from site advancing to lease executing. This is the main opportunity ahead of us to accelerate permitting, execute leases, secure our expansion capacity, and ultimately deliver to our customers. This is how we will create value for our shareholders and bondholders. Turning to slide eight.

Our execution plan is defined by six key areas, each supporting our ability to deliver at the pace and scale our future customers require. First, we've secured our deep bench of talent by adding over 60 years of infrastructure and development and over 50 years of data center construction experience combined in just the past few months. People who have delivered at scale for the most demanding customers in the world. Jonathan Mir joined as CFO, bringing 25 years of energy infrastructure strategy and project finance expertise. We have also added an SVP of construction and of power, a VP of HPC operations, and a head of permitting to oversee the execution of these critical functions. We've assembled the right team to execute on our vision. Second, we are engaging the right industry leaders as partners. T5, Turner Construction, Corgan, CWT, Vertiv.

These firms have built data centers for the world's largest hyperscalers, not once, but hundreds of times. When customers look at our project partners, which will be available on the new website when it launches tomorrow, they will see that we have also assembled the right partners to ensure better outcomes. Third, we have the capital required to bring our sites to market. As of March 27th, 2026, our liquidity stands at $520 million in cash and Bitcoin, which we expect is much more than the CapEx budgeted to get us to a lease at Panther Creek, Sharon, and Washington. Jonathan will go into more detail on our capital position and financing strategy shortly, but the headline is simple. We're well-funded and can move fast. Fourth, a disciplined Bitcoin exit. It is clear we are no longer a Bitcoin miner.

However, with strong, robust liquidity, we can have a disciplined approach to our exit strategy. We will continue to operate up until the time sites need to be prepared for construction, maximizing free cash flow before selling the miners. We will also opportunistically sell Bitcoin into strength to capture and reinvest every dollar we can into HPC and AI infrastructure. Fifth, power assets that cannot be replicated. Our megawatts sit in regions with large barriers to entry. Pennsylvania, Washington State, and Quebec all have multi-year wait lists. No one is cutting the line. Our 350 MW at Panther Creek, 110 MW at Sharon, and 18 MW in Washington were secured before the AI demand wave made these markets highly coveted.

This isn't power others can easily replicate, giving us a competitive edge with high-quality tenants who understand these markets and are hungry for assets like ours. Which leads us to our sixth point. In this market, speed to power is what drives value. For our customers, the opportunity cost of delayed deployment is huge, so the priority is getting capacity online as quickly as possible. Every day of delay is lost revenue. As a result, power availability and certainty of delivery are the primary drivers of lease economics. This dynamic has pushed lease rates higher since our Q3 call, exactly as we said it would. The opportunity in front of Keel Infrastructure is real. We now have the assets and the team is ready. I am so proud of what we built in 2025, and I'm confident in what we'll deliver in 2026 and 2027. With that, I'll turn the call over to Jonathan.

Thanks, Ben. Turning to slide nine. I joined the team five months ago. My focus has been on sharpening our approach to capital allocation, strengthening our balance sheet and capital structure, and ensuring the financing actions support long-term shareholder value creation. I've had a front row of the depth of talent, the operational discipline, and the strategic momentum across Bitfarms. I work closely with our operations and development teams, both to understand the current trajectory of our assets and to ensure our capital plans are aligned with the opportunities ahead. What stood out to me is the extraordinary potential we have, driven by the quality and potential of our sites, a strong balance sheet, the best liquidity position in the company's history, and a broad team that's both deeply engaged and committed to excellence. We're moving quickly and with purpose.

I'm pleased to be here with you today and discuss the progress we're making. I'll use this time to walk through our performance for fiscal year 2025 and outline our current capital strategy that we believe supports the accretive growth we are targeting for 2026 and beyond. Turning to slide 10. Before discussing our financials for the quarter, I want to briefly frame the results are presented this quarter. As of Q3 2025, the Paso Pe facility in Paraguay has been classified as held for sale. As a result, all revenues, operating costs, and asset balances associated with Paso Pe are treated as discontinued operations in our fiscal year 2025 financials. When I refer to continuing operations, I am speaking exclusively about our North American platform, the foundation of our transition into HPC and AI infrastructure.

With that, revenue for fiscal year 2025 was $229 million, up 72% year-over-year. Operating loss for fiscal year 2025 was $150 million, including non-cash depreciation of $98 million and $28 million of impairment charges. This compares to an operating loss of $28 million in 2024, which included $102 million of non-cash depreciation and $4 million of impairment charges. Net loss for 2025 was $209 million, or $0.38 loss per basic and diluted share, compared to a 2024 net loss of $7 million or $0.02 loss per basic and diluted share.

The differences between 2024 and 2025 were driven by a number of factors, including change in fair market value of digital assets, primarily due to the decline of Bitcoin prices and realization of gains on disposal of Bitcoin during the year. Two additional items also impacted year-over-year comparability. First, we saw a loss of $68 million, reflecting changes in our derivative assets and liabilities. Second, 2025 impairment charges were $25 million higher than in 2024. For the year, our adjusted EBITDA was $29 million compared to $31 million in 2024. Turning to slide 11. 2025 was a deliberate year of balance sheet optimization and improvement, providing the foundation for our next phase of growth.

We successfully issued an oversubscribed $588 million convertible offering, significantly expanding our liquidity. In February, we repaid the Macquarie debt facility, eliminating legacy debt, simplifying our capital structure and freeing the company from covenants. Each of these supports the pursuit of our HPC infrastructure strategy. The Macquarie facility had been originally used to accelerate development at Panther Creek, funding critical project activities, including long lead time item procurement and substation work. Retiring the facility was a strategic decision, strengthens the balance sheet and gives us the flexibility to secure more cost-effective financing at either the parent or project level.

Our current cash position of $520 million provides the runway to advance Panther Creek, Sharon and Moses Lake through lease execution without accessing capital markets, though we may do so if attractive opportunities arise that improve our ability to deliver the best possible long-term risk-adjusted shareholder returns. Macquarie was an excellent partner, and we appreciate their support so early in our pivot to HPC AI infrastructure. Turning to slide 12. As we pivot to commercialization of our development sites, we have a clear financial strategy based on three principles: capital allocation, capital formation, and capital structure. Taken together, they are designed to deliver the best possible long-term risk-adjusted shareholder returns. First, capital allocation. We deploy capital into projects where the earnings potential exceeds their weighted average cost of capital.

We rotate capital from businesses that are non-core or earning less than optimal returns and deploy the capital into higher return investments. Second, capital formation. Our financing strategy is designed to fund our very large growth opportunities while maintaining the liquidity needed for a stable base of operations. We will be opportunistic in our financing execution. We will fund construction of our data center projects using project or parent level debt and project or parent level equity or equity-linked offerings. We're taking a disciplined approach and at this time are well capitalized to actively commercialize and execute leases across Panther Creek, Sharon and Washington. Third, capital structure. Our capital structure is designed to capture the best possible long-term risk-adjusted shareholder returns while also retaining overall corporate flexibility and support of growth.

Our objective is to operate with a deliberate liquidity strategy in order to enable clear-headed commercial decisions and capital allocation decisions rather than having liquidity drive timelines. Stepping back, our roadmap is clear. We are building a regionally focused, high growth HPC AI infrastructure platform grounded in disciplined capital allocation, a strengthened balance sheet, and a development cadence that maximizes returns and minimizes risk. We're funded through the key de-risking stages, permitting and leasing across Moses Lake, Sharon, and Panther Creek. We're entering 2026 with momentum, optionality, and a balance sheet engineered for growth. We have the right people, assets, liquidity, and strategy, and we're well-positioned to capture for our shareholders the long-term value potential we have today. With that, I'd like to return the call to Ben for closing remarks.

Thanks, Jonathan. A little over a year ago, as our team began actively integrating AI into both our business and our daily lives, we came to a realization. This isn't just another technology cycle. It's a paradigm shift, more comparable to the Industrial Revolution than the Internet revolution. The fundamental measure of productivity capacity is no longer calories or joules, but tokens. This became strikingly clear two weeks ago at NVIDIA GTC, where I witnessed hundreds of companies applying AI to everything from straightforward tasks like cleaning and image generation to extraordinary complex applications, including protein folding, physics simulations, and even brain surgery. Walking the conference floor, speaking to the attendees, one thing was unmistakable: We've only begun to scratch the surface of AI's potential. Yet even in these early days, AI is already empowering individuals, communities, and companies to accomplish exponentially more.

We're witnessing Jevons paradox unfold simultaneously across every industry, thanks to AI, where improved efficiency can paradoxically drive higher, not lower, demand. It has literally never cost less to transform an idea into an action, a product, an image, a refined concept, a service, or countless other outputs. The possibilities are truly limitless. While no one can predict exactly how AI will reshape our future, one certainty remains. It will require enormous amounts of power. Our 2.2 GW of capacity and strategically positioned land across Pennsylvania, Washington, and Quebec sit directly in the path of this transformation, and we intend to capitalize on that opportunity for our shareholders, and we look forward to the opportunities ahead. With that, I would like to open the call to Q&A. Operator, please go ahead.

Thank you. One moment for our first question. Our first question comes from Mike Grondahl with Northland. Your line is open.

Hey, thanks, guys. First question, Ben. You talked about your decision not to go the GPU rental route at Moses Lake, and just the colocation route. Could you talk a little bit about what a couple of the major drivers were that got you to that decision?

Yeah, it's a great question, Mike. You know, when we first started talking about in Q3, we were always evaluating this alongside with colocation. You know, we're trying to maximize the value for shareholders, so we're always going to evaluate multiple different business models at our sites. Because, you know, they have the lowest cost energy and all these other benefits, we thought it would make a lot of sense. As we've continued to have increasing amounts of customer conversations for Washington and other sites, it was just really clear to us that the best opportunity for us is to just remain a pure play infrastructure developer and owner and let these customers who really want these megawatts lease these megawatts.

Got it. Maybe secondly, you articulated, I'll say, a philosophy a quarter or two ago about waiting on signing a lease as terms were continuing to improve, you know, kind of implying you were gonna be really patient and wait on a lease. Could you kind of update how you're thinking about that lease execution strategy and the potential timing around it?

Yeah. You know, our strategy on lease execution has been consistent. It remains consistent today. You know, our view is that the best way to maximize value for shareholders is to get the best terms in a lease, because that's gonna be what is gonna be driving our NOI and our multiple. When we're looking to sign 10 to 15 year agreements, you know, it's really important for us to take the, you know, maybe a little bit more time than investors may want us to in order to get better terms for longer. When it looks at what is really driving the value in these lease economics, one of the biggest elements is risk. We've spoken to this, you know, multiple times over the last couple of months.

The you know biggest risk for most of these sites is. It's possible to go out there and, you know, have conversations and get a lot of interest. In some cases, you could even sign a lease prior to getting permits. All of that risk is gonna be priced into the agreement. You're gonna be locked into it for 10-15 years, and that's gonna negatively impact, you know, the long-term value that we're creating for shareholders. Our strategy has been incredibly consistent.

The benefit for us is that, you know, we are operating in high-demand markets with high barrier entry, so it takes a little bit longer to get permits going in Pennsylvania or in Washington than it does in Texas, which is the easiest market in the United States for that. We believe that drives a lot of extra value because it's way more scarce, it's way harder to acquire, and there's just not as much optionality.

Got it. Well, thanks. Hey, good luck in 2026.

Thanks, Mike.

Thank you. Our next question comes from Brett Knoblauch with Cantor Fitzgerald. Your line is open.

Hi, guys. Thanks for taking my questions. Maybe to start, could you maybe just go into detail on what permits at what sites you guys are waiting to receive?

You know, permits is a complicated process, and we're getting permits across multiple sites in multiple jurisdictions. They all have different rules, different regulations, different timelines, different reviews, different, you know, authorities. You know, it's far too much detail to get into exactly what permits are remaining on all the different sites. We are continuing to make good progress. Kind of, you know, we're looking at the visibility over the next couple of months, and with what we've had so far with the community engagement success that we've had so far, we think that, you know, in the coming months, sometime around the mid to late summertime, we should be achieving the full permitted status across at least one, if not all of the sites.

Perfect. Maybe just on the leasing environment across the different sites that you guys have. I guess we were under the impression that maybe Sharon would be, you know, first to go, given it's relatively further along. Is that still how you guys are thinking about it? In the presentation, when you guys kind of list, you know, the power pipeline and roadmap, how much of that is from generation on site that you guys are looking into? Do you have any update on where you guys are with respect to, you know, sourcing that generation?

Yeah, sure. To answer the second part of your question first, you know, all the power that we're talking about developing for our HPC and AI data centers right now is grid connected. The two operating power plants that we have at Scrubgrass and Panther Creek currently that math is not in those charts for the secured capacity or the site development plans. But in Scrubgrass particular, we are working to expand the generation capacity there with natural gas. We've been working to tap into the Tennessee Natural Gas Pipeline. We're achieving, you know, pretty good results there with the engineering firms. There's still probably another month or two to go before we're getting, you know, a clear path forward on the engineering plans. You know, Scrubgrass is more of our pipeline site, and so that power generation opportunity is more of a 2028 and 2029 timeline. Everything else is grid connected, it's secured today, or it's currently active. Sorry, Brett, I'm blanking on the first part of your question. Would you mind repeating it?

Yeah, just on maybe the cadence of which sites are maybe quicker to go. Yeah.

Yeah. Really that's gonna be driven by success on permitting timelines and the customers. All three of the sites, Moses Lake, Sharon, and Panther Creek, are all actively in our go-to-market right now. Every single one of those has customers engaged under MNDA, and they have for quite some time. We're continuing to push forward on those conversations and those negotiations. Really, I think what investors should think about with regards to permits are more of a closing condition to a lease, right? They're really not a starting condition to a negotiation. We have these conversations and these negotiations simultaneously while we're working towards permitting. As permitting gets closer and closer, the negotiations will also get closer and closer in tandem, and the first site to get leased is likely to be the first site to be permitted.

Awesome. Really appreciate it. Thank you, guys.

Thanks, Brett.

Thank you. Our next question comes from Stephen Glagola with KBW. Your line is open.

Hey, thanks for the question. Just on that last point, if you could clarify the sequencing here between like notice to proceed and lease execution. In other words, like, can you pre-sign leases contingent on notice to proceed, or is like notice to proceed required before any major customer would commit to a lease?

For a customer to commit to binding, in our view, they're going to want NTP, and that's based on the number of conversations that we, you know, continuing to have. There probably are some customers who would be interested to sign prior to NTP, but those aren't the investment-grade counterparties that we're really seeking to engage with.

Okay. Thank you. Just one more. You know, how are you thinking about like Vera Rubin hardware availability in 2026 and like early 2027, and to what extent could that variability in supply influence the timing, you know, lease discussions at your sites? Thank you, Ben.

Yeah, it's a good question, Stephen. You know, we've been talking about Vera Rubin, I think, since Q3 call, because all of our sites, you know, are basically coming online in 2027, so we're trying to make sure that they are designed for the highest level of equipment that's coming out in 2027 and 2028, which is the Vera Rubins. In terms of supply, we haven't seen any, you know, impact so far. I understand, you know, there's always geopolitical uncertainty in the world that may impact those supply chains. But given that energy is such a huge bottleneck, and it's always been the huge bottleneck on the growth. I don't think that there's going to be a geopolitical situation that's going to make the bottleneck change from energy over to GPUs. We don't have any expectation right now that that's going to have any impact on leasing or demand for sites because power is still such an extreme bottleneck. It's hard to imagine what's going to overshadow that geopolitically.

Great. Thanks, Ben.

Thanks, Stephen.

Thank you. Our next question comes from Michael Donovan with Compass Point. Your line is open.

Hi. Thanks for taking my question, and congrats on the progress. Can you provide an update on ESA progress, specifically Panther Creek's ISA to ESA conversion?

Yeah. That's a great question, Mike. You know, as investors probably know, we have 350 MW secured ESA with PPL. In addition to that, we also have an ISA that enables us to draw down approximately 60 MW from the grid, and that's associated with the existing transmission line and substation for the power plant that we currently have operating. In order to get that converted over, it's really more of a regulatory matter. It's hard to put an exact timeline as to, you know, when those stamps are gonna be received. There's no, you know, infrastructure that needs to be built. There's no CapEx that needs to be spent.

Really, it's just a matter of getting the regulatory approval to convert a non-firm service into a firm service, and that would enable us to increase our capacity beyond 350 MW to what we probably expect is gonna be maybe 400 MW or possibly slightly more. We expect this is gonna happen this year, but it's hard to put an exact timeline on it given it's a regulatory matter.

Great. Appreciate that, Ben.

Thanks, Mike.

Thank you. Our next question comes from Brian Kinstlinger with A.G.P. Your line is open.

Great. Thanks. Last quarter, Ben, you communicated you expected the GPU as a service at Moses Lake site would be targeted for, I believe, the first quarter for go live. How does shifting to colocation change the timing, if at all? And my second question is, can you talk about also how the global memory shortage is impacting your site development or changing your near-term needs or planning for lead times?

Yeah. Two parts to that question. In terms of switching from a GPU as a service to colocation, just changing the business model doesn't really impact the development timelines. We don't really see any delay there associated with changing from GPU as a service just to colocation. Really, it's just a matter of how we want to allocate our capital and how we wanna focus the business. When it comes to the memory shortage, you know, as a pure play infrastructure developer and owner, that really is not coming into our calculus very much. Mostly that's a customer situation for them to resolve with their own supply chain because we're not the ones investing in the GPUs and the compute and the servers.

Great. Thank you.

Thank you. Our next question comes from Martin Toner with ATB Capital Markets. Your line is open.

Good morning.

Yeah.

Question. Can you guys elaborate on to increase capacity beyond. Can you kinda give us some timelines, thoughts there?

I'm gonna repeat the question 'cause it was a little quiet, just in case nobody else or other people had difficulty hearing. I believe the question was, can you give some timelines as to how we might be able to expand Panther Creek to 500 MW and beyond? In order for us to move beyond the 350 MW ESA that we have secured, there's really two sources for expansion. The first is converting over that ISA from non-firm service to firm service that I just spoke to a minute ago. That's really a regulatory matter that we expect to resolve sometime this year. It could be tomorrow, you know, a few months from now.

When it comes to expanding beyond that, what we have to do with that is we have to actually have new power applications. The good thing here is that the utilities are actually looking to invest in new generation in the area. In this particular instance, we weren't actually applying for new power. We actually had the utility call us and ask us how much more power we could take on site. You know, given the bottleneck constraint on power, that was obviously a very welcome call over here at Bitfarms to receive. It's a pretty unusual one in the industry. They're looking to scale up generation capacity in the area, specifically to service our site at greater capacity.

This is probably going to be two to three years timeline 'cause there's a lot of process involved with spinning up new generation and building those new transmission lines. For a lot of our customers, what they really want is, you know, the fastest pathway to energization and a clear path to scale over multiple years. This really lines up with what the hyperscalers and what the Neoclouds are searching for.

That's great. Thanks very much. Hopefully, you can hear me better. Can you clarify when you expect to sign your first lease?

You know, I can't get into a specific timeline, but in terms of milestones, as I spoke to earlier, it's really about clearing NTP as kind of the last closing condition or last milestone for us to sign a lease. I think for the investors and the analysts on the call, the important thing to keep track of, especially over the next coming months, is the continued progress that we have towards NTP. Because once NTP is cleared, that's basically the last thing standing between us and a signed agreement.

Got it. Great. Thanks. Last one from me. Can you talk a little bit about why your mining exahash in Q4 was at the level that it was at?

We continue to scale back our mining exposure as we continue to focus on our U.S. HPC infrastructure investments. You know, we haven't made any investments into Bitcoin mining. We're not spending any money on upgrades or new miners, and we're actively working to scale down the fleet and actively working to spin off assets like we have in Paraguay that are not suitable for conversion. Investors should continue to expect our hash rate to continue to trickle down over 2026 as we continue to execute on this transition to HPC and AI.

Thank you very much, Ben.

Thanks, Martin.

Thank you. Our next question comes from Mike Colonnese with H.C. Wainwright & Co. Your line is open.

Hi. Good morning, Ben and team. Thank you for taking my question this morning. Ben, I'm just curious, after securing the remaining permits across the three sites, which sounds like will likely take place in the coming months here, what does the timeline look like from a data center construction and delivery standpoint? It sounds like you're pretty optimistic that revenue generation could commence as soon as next year, but any additional color here would be helpful.

Yeah, I mean, really this is the year of execution, and 2027 is the year of delivery. At all three of our projects that we talked about today, Panther Creek, Sharon and Washington, we all expect them to come online and start delivering megawatts and start generating revenue to customers in 2027. We'll continue to provide updates as we go along. I think once we have cleared NTP and we have signed leases, there's gonna be a lot clearer visibility that we can provide to investors for each specific project and their specific timelines.

Got it. Thanks for that. Back to Bitcoin mining operations. It sounds like you're progressively gonna be scaling back hash rate as you bring some of the HPC AI data centers online. I guess what's the best way to think about, you know, hash coming offline and kind of flowing through your operating results within the near term here?

Well, I'll speak to it at a high level then, and then maybe I'll pass it off to Jonathan for some further clarity. You know, right now the Bitcoin mining remains profitable, but it's not very. It's marginal. It's still contributing to the business, but really it's not the focus of the business. It's not where we're investing our time. It's not where we're investing our efforts. Given that we have been so successful last year in raising capital and strengthening our balance sheet, it's really not super impactful for the developments that we have this year, the operations or the CapEx. We'll just continue to scale that down, trying to maximize value in the disciplined exit.

You know, if it makes more sense to maybe sell some miners a little bit earlier than we might need to in order to begin construction, we'll evaluate that as we will, you know, always do to maximize value for our shareholders. Really we kind of see this as, you know, a pretty minor element of our balance sheet and a minor element of the financial plan for this year. Jonathan, do you wanna add anything further?

Only that when we think about our liquidity going forward, the strategic objective is to ensure we are well capitalized through the lease process and beyond, without the need to raise any new capital in the markets. That takes into account the current state of Bitcoin mining operations. It's not assuming any improvement in the economics there. Our plan is built on conservative assumptions around the status of the Bitcoin market.

Very helpful. Thank you for taking my questions.

Thanks, Mike.

Thank you. Our next question comes from Nick Giles with B. Riley Securities. Your line is open.

Hey, good morning, Keel team. You know, in the interim period where Bitcoin mining operations are wound down, but kind of pre-revenue generation on the HPC side, could the generating assets at Panther Creek and Scrubgrass be utilized in any way, such as the PJM capacity auction?

Those power plants do actually participate in PJM capacity auctions. We've done that for quite some time. We do benefit from the capacity payments that we receive there.

Got it. Okay. Any order of magnitude of what those could be kind of in the 2026 planning year?

I mean, really, we've kind of maxed out on the capacity auction payments. They set a ceiling, and that's where the capacity auction payments closed.

Got it. Understood. Maybe one for Jonathan. You know, you've made some progress on the capital structure, but just was hoping for any additional comments you might have on what you're looking for in you know, an initial debt package, how you're seeing terms shift and kinda what tools you'll have at your disposal during construction and kinda post-energization.

Good question. Thanks, Nick. Our basic approach is to compare and contrast our financing options down at the asset level and upstairs at the parent level. Certainly one of the things that we've seen in the market that has caught our attention like everyone else is the tightening of spreads between folks issuing high-yield debt in the market at what seem like you know quite attractive levels for strong investment grade counterparties or credit wraps. Those converging towards the level seen in the bank-originated classic construction and project financing. Each of those has its own advantages in terms of you know simplicity of managing the actual capital once it's raised versus negative carry costs. As we get closer to a funding point, we'll make a decision as to what seems best for our shareholders in terms of how we decide to finance. Right now what I would say

Great. Thanks, guys.

Oh, I'm sorry, Nick. I was just gonna say that the markets for our space and for infrastructure generally seem calm right now.

Understood.

Thank you. Our next question comes from Brian Dobson with Clear Street. Your line is open.

Hi, it's Craig Kendy in for Brian Dobson. I guess one final one. Just on the re-domiciliation filing to the U.S., are there any implications to costs or, you know, structural implications in terms of ownership that we should be aware of as you enter this over the next couple of days? Thanks.

Good morning, Brian. One of the benefits and reasons for the redom is that we will now be eligible for inclusion in indices that require one to be a U.S.-domiciled company. For example, we'll be eligible for inclusion in the Russell 1000 and the Russell 3000, as well as for ownership in any other fund who was otherwise limited to the purchase of U.S. securities. We view that as being, you know, quite helpful in terms of moving our shareholder base to one that is institutional and long term. There are no cost or flexibility implications on our end. We simply see this as a nice path forward with a lot of benefits for our shareholders.

Very helpful. Thanks a lot.

Thank you. Our next question comes from Bill Papanastasiou with Chardan Capital Markets. Your line is open.

Yeah, good morning. Thanks for taking my questions, gentlemen. Just wanted to touch on the Washington site and decision to shift towards colo. Can you confirm that this won't have any material impact on the purchase commitment that was entered into November? Or, you know, is the team considering the shift in development, allocation to other sites?

Thanks, Bill. No impact on the capital commitments and the equipment we've already purchased for the Washington site by changing business models. In fact, yeah, actually, Bill.

Awesome.

It just helps to reduce the CapEx 'cause we're no longer paying for the compute.

Understood. Thanks. How should we generally be thinking about maintenance CapEx on existing Bitcoin mining sites as you gradually shift over to AI HPC here?

We're not making any investments into the Bitcoin mining sites. Basically, we're just continuing to keep them up and running. No further investments are being made in the sites, into new sites, or into new miners.

Great. Appreciate the color. Thank you.

Thank you, Bill.

Thank you. This concludes the question and answer session. I'd like to turn the call back over to Ben Gagnon for closing remarks.

Thank you very much, everyone, for joining our call today, and really look forward to speaking to you next time as Keel Infrastructure. Have a great day.

Thank you for your participation. This does conclude the program. You may now disconnect.